Brachytherapy (BT) is advantageous due to high absorbed dose conformity and possibility to deliver high dose in few fractions. It is often used for prostate and gynecological tumors as monotherapy or as a boost alongside external beam radiotherapy (EBRT). However, a number of things can compromise treatment delivery, starting from incorrect source data in a treatment planning system to malfunctioning of a treatment delivery unit. The established quality assurance (QA) covers individual aspects, such as software checks of absorbed dose calculations, mechanical checks, source dosimetry. None of them emulate treatment delivery where the planned dose could be compared with the experimentally determined values. While such practices are employed in EBRT, BT suffers from the lack of detectors that would be water-equivalent and convenient to use for regular measurements. First-choice thermoluminescence dosimeters are water-equivalent but have passive readout. Sporadic attempts to use other detectors have not led to any established practices at clinical sites. Stepping ahead, the safety of treatment delivery could be further evaluated using real-time in vivo dosimetry. If detectors were characterized with high accuracy, a reliable error detection level could be set to terminate treatments if needed. Contrary to the inphantom QA, there are detectors suitable for such applications, but their characterization is incomplete. In this thesis we address both problems.Focusing on high-dose-rate 192 Ir remote afterloading treatments, which are among the most common in BT, we investigate and propose a direct readout synthetic diamond detector for the in-phantom QA of treatment units. The detector was designed for small-field highenergy EBRT dosimetry, but our findings demonstrate its suitability for BT dosimetry. Additionally, due to the detector calibration with traceability to primary standards of absorbed dose to water of high-energy EBRT and a combination of experimental and Monte Carlo (MC) characterization, the uncertainties in the determined absorbed dose to water values were comparable to or lower than for other detectors used in BT. We complemented the investigation with a theoretical study on diamond material properties and which values (mass density, mean excitation energy, number of conduction electrons per atom) shall be used for the most faithful description of ionizing radiation interactions in diamond for MC simulations and calculations of mass electronic stopping power. The findings improve diamond dosimetry accuracy not only in BT but also in EBRT where the detectors are used.